Nelson Pass said:
The article said the enclosures used were sealed enclosures. Sealed enclosures won't reduce the driver's mecahnical Q, which is what Thorsten has been talking about.
se
Steve Eddy said:
The article said most of the enclosures were sealed. They still
exhibit considerably less Q due to acoustic resistance, as
seen in the impedance curves which show both cases on the
same graph. In addition, the small amount of electrical damping
which is allowed dramatically lowers the Q further, due mostly
to the very light moving assembly (voice coil and cone) and the
high quality motor.
Nelson Pass said:
Yeah, but with source impedances around 47 ohms, and driver impedances at resonance in sealed boxes typically ranging from 50 to 90 ohms, isn't it rather a stretch to call this current drive?
Sure. These things weren't designed for use on open baffles or infinite baffle enclosures. But yeah, if you want to use them this way and bring up the low frequency response, increasing the source impedance in order to reduce the amount of electrical damping is one way to do it.
se
Konnichiwa,
Steve, you seriously disappoint me. As you (should) well know, both FETs and Pentodes are in effect transconductance devices, or in other words, modulated (nonideal) current sources. As such they invariably have a very high Anode/Drain/Collector impedance, compared to the load. A common Audio pentode loaded with a 2.5K SE Load may very well have 10 - 20K Anode impedance. Hence it behaves like a non-ideal current source, at least when compared to a triode driving the transformer/speaker combo.
On the contrary, if you had read Nelsons paper you would realise that the EXPLICIT aim is to provide a system with a near critical damping. No, the aim is not a Q=0.7 System most of the time, as many people perfer (as also Nelson remarked and was observed by Jecklin and others) a Q around 1 - 1.2 (for whatever reasons).
My original point was that by using the right measures (of which I know of around 4 or 5 and can imagine several more) you can by making the Amplifier, Driver and Enclosure (or lack thereoff - Dipole) a "tuned system" achieve any sensible desired LF response efficiently and with good transient response.
Of course, if you do, there is no room for an Amplifier that can drive any speaker (not that there ever was, I have yet to hear an amplifier that is equally good with any speaker, thopugh most amplifiers are equally bad no matter what the speaker) and a speaker attachable to any Amplifier (see corollary earlier - there is no such thing anyway).
So, we are back in the realm of the old big Valve Radio and Music Center, where enclosure, Speakers and Amplifier where threated as ONE system and optimised to play together, in direct and polar opposition to current HiFi & High End practice.
Now, if you hear what the guys in the old days achieved in musical terms with systems EXTREMELY compromised by economics one shall have to wonder what a "cost is only a moderate object" implementation of such principles can achieve. And if I am not mistaken what prompted Nelson on his "first watt" path was to attempt to answer this question, right?
Sayonara
Steve Eddy said:
Steve, you seriously disappoint me. As you (should) well know, both FETs and Pentodes are in effect transconductance devices, or in other words, modulated (nonideal) current sources. As such they invariably have a very high Anode/Drain/Collector impedance, compared to the load. A common Audio pentode loaded with a 2.5K SE Load may very well have 10 - 20K Anode impedance. Hence it behaves like a non-ideal current source, at least when compared to a triode driving the transformer/speaker combo.
Steve Eddy said:
On the contrary, if you had read Nelsons paper you would realise that the EXPLICIT aim is to provide a system with a near critical damping. No, the aim is not a Q=0.7 System most of the time, as many people perfer (as also Nelson remarked and was observed by Jecklin and others) a Q around 1 - 1.2 (for whatever reasons).
My original point was that by using the right measures (of which I know of around 4 or 5 and can imagine several more) you can by making the Amplifier, Driver and Enclosure (or lack thereoff - Dipole) a "tuned system" achieve any sensible desired LF response efficiently and with good transient response.
Of course, if you do, there is no room for an Amplifier that can drive any speaker (not that there ever was, I have yet to hear an amplifier that is equally good with any speaker, thopugh most amplifiers are equally bad no matter what the speaker) and a speaker attachable to any Amplifier (see corollary earlier - there is no such thing anyway).
So, we are back in the realm of the old big Valve Radio and Music Center, where enclosure, Speakers and Amplifier where threated as ONE system and optimised to play together, in direct and polar opposition to current HiFi & High End practice.
Now, if you hear what the guys in the old days achieved in musical terms with systems EXTREMELY compromised by economics one shall have to wonder what a "cost is only a moderate object" implementation of such principles can achieve. And if I am not mistaken what prompted Nelson on his "first watt" path was to attempt to answer this question, right?
Sayonara
Nelson Pass said:
Sorry for the confusion. When I said "enclosures" I was referring to the sealed enclosures used for the majority of the drivers. I'd commented that the article didn't address the high mechanical Qs of the drivers. You responded by saying "It was addressed in an actual enclosure, where it really counts."
Since the Mini Medallion, the Abbey and the KleinHorn are conventional enclosures and originally designed and intended to be used with voltage source amplifiers, and since the KleinHorn appears to have been designed no differently, I assumed that when you said that the high mechanical Q was addressed in an actual enclosure, I assumed you meant the various sealed enclosures used on the other drivers.
Which Q are you refering to here?
I can't say I've seen an example where a sealed enclosure has produced a system Q lower than that of the driver's total Q.
Well sure. You're just allowing more of the damping that's largely responsible for the low driver Q which you want to increase in the first place. 🙂
se
Kuei Yang Wang said:
Sorry about that, chief. 🙂
Yes, I'm aware of that.
Yes, I'm aware of that too.
Yes, but that 10-20k anode impedance is effectively in parallel with the 2.5k load impedance isn't it? So that the output impedance would be around 2k ohms, yes?
Didn't those old amplifiers use output transformer as well in order to reduce the output impedance of the amp to something closer to the loudspeaker's impedance?
Vintage tube amps from the 30s, 40s and 50s aren't my forte but of the ones I have seen didn't have output impedances on the order of thousands of ohms.
Can you point me to some examples of these?
Ok. Just that most people seem concerned about low frequency transient response and try to eliminate overshoot and ringing.
Sure.
Agreed.
Quite so. But then "optimized" can have more than one meaning. You can optimize such a system for some singular objective techincal goal. Or you can optimize such a system to suit the tastes and preferences of a particular individual.
Just as you say above that there was a singular amplifier to drive all speakers, neither is there a singular system that will satisfy all listeners.
Well if First Watt is geared toward ultimately producing a singular "optimized" system, things would make a bit more sense. 🙂
se
That doesn't make sense to me. When I look at that circuit, it looks like, if the secondary is an 8 ohm winding, that the amp will have something like a 50-60 ohm source impedance. You know transformers pretty well, better than me, I think, so you'll pardon me saying simple stuff. The 2.5K:8 is the stepdown ratio of the source impedance of the pentode. The 2.5K isn't an actual resistance, it's just the impedance that the transformer provides as a load to the plate if there's 8 ohms on the other side. So the transformer makes the speaker look like 2.5K to the tube. Or looking at it the other way, the transformer makes the tube look like 50-60 ohms to the speaker.
SY said:
I wasn't entirely sure that there was even a trasnformer in the scenario Thorsten mentioned when he said "Hence it behaves like a non-ideal current source, at least when compared to a triode driving the transformer/speaker combo."
I guess he did mean a pentode with an output transformer. Otherwise, I was thinking of something along the lines of a common-emitter with a 2.5k collector resistor seeing as tubes ain't my bailiwick. 🙂
In any case, whether pentode or triode, weren't the transformers wound so as to ultimately have an output impedance roughly that of the loudspeaker's impedance? If so, I still think it's quite a stretch to call such an output a current source, especially considering that the issue at hand here concerns the driver at resonance, where its electrical impedance may be as high as 90 ohms or more.
I mean, where exactly do you draw the line between voltage source with a rather high output impedance and a current source with a rather low output impedance?
se
Steve Eddy said:
I don't think we need to draw a line, rather recognize that
the ideal source impedance is between 0 and infinity. If we want
0, then a perfect voltage amp is called for, and when we want
infinity a perfect current source is what we want.
When the desired source impedance is greater than that of the
speaker, I think we're probably looking for a current source with
some shunt value. When it is less, we probably want a voltage
source with a series value (if any). We are not, of course,
required to use one or the other, but it does get silly putting
50 ohms in series with the output of a Krell, or a 2 ohm resistor
in parallel with an F1.
Nelson Pass said:
Too limiting. There's the negative resistances, too. And why limit yourself to real values? Why not think about an amp with a reactive source impedance?
Nelson Pass said:
Ok, but here it just ends up being a matter of bumping up the output impedance a bit in order to get the Q up rather than what one thinks of when they think of driving speakers with current sources.
But at your primary areas of interest, the bottom end to bring up the driver's Q and the top end to bring up treble response, the load impedance is roughly that of your source impedance and in some cases even higher.
Why not address the driver's mechanical Q as Thorsten suggests and then use a REAL current source instead of what amounts to little more than a voltage source that wants to grow up and be a current source some day?
Because you're left with rather high mechanical Qs, you're stuck with these really low output impedances (really low for a current source) and don't get but a small fraction of the potential benefits of current drive.
I'd suggest you get together with Thorsten and work something up to address the mechanical Q issue but he seems more interested in securing patents than anything else. 🙂
Sure.
se
Konnichiwa,
You know, I was sorely tempted to call a few names and get myself sinbinned for a while.
There are two options, either you really don't know what you are writing about, in which case I suggest you read up a little on the subject and make some experiemnts of your own, or you insist on arguing for arguments and semantiks sake, despite knowing well that the position you present is silly.
Nope. You really don't seem to get it? Simple, a transformer is a transformer. If we assume an ideal one, the load (speaker) when connected to the transformer is simply "geared up" without losses. So a 2k5:8 transformer simply makes an 8 ohm speaker appear as 2k5 load in the anode circuit of the output valve (pentode).
This valve in turn has a certain anode impedance, which is in effect the non-ideal behaviour of our non-ideal current source.
If we operate in the Valve circuitry area we treat the triode as non-ideal voltage source (in other words a voltage soure with a given series resistance) while we usually treat tetrodes and pentodes as non-ideal current sources, meaning a current source with a parallel resistor. The same actually applies to transistors and FET's, except they do not have a "triode" equivalent, they are in effect all non-ideal current sources in the output loop.
If we use sufficient (loop) negative feedback we can make any of the devioces present as output stage in an amplifier either a very low or a very high output impedance, making less non-ideal voltage or current sources. But some people prefer to manage without loop feedback, which after all can only correct an error after it has already occoured.
And please note that I used the term NONIDEAL CURRENT SOURCE.
I don't object to such a design goal. I merely object to first maximising the resonances and ringing in the acooustical (mechanical) system and to then hope to "control" this by shortcircuiting the back EMF. I merely suggest to arrange the acoustical system such that acoustical system has the desired transient, step and frequency response regardless of source impedance. Or in other words, to put the horse before the cart and not the other way around.
To conclude I feel your writing in this thread is simply w***ing around as you insist to disagree. If you cannot find anything remotely factual you switch to semantics. The key reason for your disagreeament however is sadly profound ignorance of the subject, so I shall save myself further discussions untill and when you have aquired sufficient fundamental knowledge of electronics and acoustics to actually hold a sensible and topical discussion.
Sayonara
You know, I was sorely tempted to call a few names and get myself sinbinned for a while.
There are two options, either you really don't know what you are writing about, in which case I suggest you read up a little on the subject and make some experiemnts of your own, or you insist on arguing for arguments and semantiks sake, despite knowing well that the position you present is silly.
Steve Eddy said:
Nope. You really don't seem to get it? Simple, a transformer is a transformer. If we assume an ideal one, the load (speaker) when connected to the transformer is simply "geared up" without losses. So a 2k5:8 transformer simply makes an 8 ohm speaker appear as 2k5 load in the anode circuit of the output valve (pentode).
This valve in turn has a certain anode impedance, which is in effect the non-ideal behaviour of our non-ideal current source.
If we operate in the Valve circuitry area we treat the triode as non-ideal voltage source (in other words a voltage soure with a given series resistance) while we usually treat tetrodes and pentodes as non-ideal current sources, meaning a current source with a parallel resistor. The same actually applies to transistors and FET's, except they do not have a "triode" equivalent, they are in effect all non-ideal current sources in the output loop.
If we use sufficient (loop) negative feedback we can make any of the devioces present as output stage in an amplifier either a very low or a very high output impedance, making less non-ideal voltage or current sources. But some people prefer to manage without loop feedback, which after all can only correct an error after it has already occoured.
And please note that I used the term NONIDEAL CURRENT SOURCE.
Steve Eddy said:
I don't object to such a design goal. I merely object to first maximising the resonances and ringing in the acooustical (mechanical) system and to then hope to "control" this by shortcircuiting the back EMF. I merely suggest to arrange the acoustical system such that acoustical system has the desired transient, step and frequency response regardless of source impedance. Or in other words, to put the horse before the cart and not the other way around.
To conclude I feel your writing in this thread is simply w***ing around as you insist to disagree. If you cannot find anything remotely factual you switch to semantics. The key reason for your disagreeament however is sadly profound ignorance of the subject, so I shall save myself further discussions untill and when you have aquired sufficient fundamental knowledge of electronics and acoustics to actually hold a sensible and topical discussion.
Sayonara
Kuei Yang Wang said:
Then there's the third option, I wasn't exactly sure what you were talking about.
Ok. I wasn't sure in your original comment whether there was a transformer in the picture or not. Your main emphasis seemed to be on the 10-20k anode impedance which seemed to imply that this was the impedance through which the speaker was being driven as you were saying they were driving these speakers with current sources.
But transformers work both ways. The same thing that causes the 8 ohm loudspeaker impedance to be seen as a 2.5k anode load by the pentode also causes the 10-20k anode impedance to be seen as a 32-64 ohm source impedance by the loudspeaker.
But at resonance, which is the area of interest here, the loudpeaker's impedance is typically going to be as high or higher than 32-64 ohms.
I'm sorry, but when the load impedance is as high or higher than the source impedance, I have a hard time thinking of this as being a current source, even a non-ideal one. To me it seems to be more a non-ideal voltage source.
Except that the loudspeaker doesn't see that certain anode impedance. Yeah, if the loudspeaker was being driven by a source impedance of 10-20k, I'd agree that this could be considered a non-ideal current source. But in the case you're talking about here, it's being driven by an impedance about 300 times less than that, which in the area of interest here may not even be as high as the loudspeaker's impedance.
And as I said above, I have a very hard time seeing this as a non-ideal current source but rather a non-ideal voltage source.
Yeah, because triodes have a lot of internal feedback and hence a much lower anode impedance compared to a pentode.
Sure, if you're just considering their drain/collector impedances. But if the subject is driving speakers with a current source, then even if they had infinite drain/collector impedances and were perfect current sources, it's irrelevant. If you're using them as a common source/emitter amplifier, the source impedance seen by the loudspeaker is going to be depend on the drain/collector load, which is effectively in parallel with the drain/collector impedance.
Yes, I've noted that. And I agree that if you were driving the loudspeaker with a 10-20k source impedance then "non-idea current source" would be apt.
Why do you say hope to control? It quite obviously does control it. Some would argue that it controls it too well, which is the whole purpose of increasing output impedance, yes? To remove some of that control?
I don't see how you could design a speaker which would have the same transient, step and frequency response regardless of source impedance. Seems to me that would be rather like designing a passive crossover which has the same response regardless of source impedance.
Unless I'm missing something, seems to me you can only optimize for either a voltage source or a current source.
"W***ing"? Wanking?
I don't insist to disagree. I disagree when I disagree and I agree when I agree. As it should be in my opinion. Should I agree with that which I disagree with just to please those who are intolerant of anyone who doesn't agree with them?
se
Steve Eddy said:
But you do have a tendency, Steve, to lead a tread astray with specious agruments... i believe you have been lectured on this a number of times....
dave/planet10
planet10 said:
Specious. Having deceptive attraction or allure. Having a false look of truth or genuineness.
I'm a lot of things, but disingenuous isn't one of them.
If that's how you see it, then I would adivse you to just ban me now and get it over with because I will continue to question that which I find questionable and disagree with that which I disagree with.
se
Resist that temptation. If there are questions that you think are silly or otherwise not worth answering, then don't answer them. That's considerably more civil.
Payin my dues
This is the promised mixed solution for combined voltage- and current- drive. I didn't try to get the flat part of the curves to the same level. One can still see the outcome like that (maybe even a little better).
The top circuit is the mixed current- and voltage- drive. Middle one is pure current-drive. And the lower one the pure voltage-drive.
The traces are: Green for the mixed solution, red for the pure current-source and blue for the pure voltage-source.
Regards
Charles
This is the promised mixed solution for combined voltage- and current- drive. I didn't try to get the flat part of the curves to the same level. One can still see the outcome like that (maybe even a little better).
The top circuit is the mixed current- and voltage- drive. Middle one is pure current-drive. And the lower one the pure voltage-drive.
The traces are: Green for the mixed solution, red for the pure current-source and blue for the pure voltage-source.
Regards
Charles
phase_accurate said:
No need. Save a few francs on your power bill. 😉 Just broadening perspectives, not seeking designs.
A power GIC could easily be designed with, for example, a source impedance that is a conjugate of the load.